Chapter 9

Boundaries of Understanding Personality-Ability Relations

CHAPTER SUMMARY

1. Our meta-analyses are orders of magnitude larger than previous investigations and illuminate novel areas of overlap between personality traits and cognitive abilities.
2. Nevertheless, we assessed potential limitations and considered opportunities for future exploration:
   1. Publication bias is unlikely to have affected this book’s major conclusions since we made every effort to include the population of available studies and since most of the effect sizes came from unpublished sources, including diverse participants from around the world.
   2. Some personality-intelligence correlations had to be estimated with limited data, because primary research has not yet comprehensively explored relations between these domains. Results based on sparse data should be interpreted as preliminary estimates and examined in future research.
   3. Linear associations were examined in all contributing research. Non-linear associations are possible, but if they exist, then the linear estimates presented here are under-estimates of their true strength. Future research should examine potential non-linear associations, especially in clinical measures and populations.
   4. Extremely large studies influenced some results. Project Talent was an outsized contributor to multiple analyses. As a nationally representative, professionally run study, its results should be an accurate representation of associations in the population, but if they are not, some relations may be biased. We encourage new, large-scale studies to further expand our knowledge.
   5. Causality and longitudinal effects cannot be inferred from our investigations since we focus on cross-sectional associations. Longitudinal studies offer some of the greatest opportunity to make significant discoveries.
   6. Unexamined moderators may influence some findings, but for many relations there was little heterogeneity across effect sizes to be explained.

The current findings highlight promising avenues for future investigation, and they also have implications for research, theory, and applications. In this chapter, we first place our results in context of other major studies of personality-cognitive ability relations. We then describe some potential limitations and boundary conditions of our research, providing directions for future research.

Interpreting Contributions of Findings

We will not rehash all the discoveries from our results here. While our research is not the first to examine personality-cognitive ability relations, it was the most detailed and comprehensive study to date.

Beyond documenting relations that are explainable by Cybernetic Trait Complexes Theory (CTCT), our illumination of “missing areas” of overlap between personality and cognitive ability are of significance for knowledge acquisition theories (e.g., Ackerman, 2018). For example, Ackerman and Heggestad’s (1997) math and spatial abilities complex did not include any personality constructs, leaving voids in understanding. The findings from our research add higher standing on the openness’ ideas facet, and lower standing on politeness and modesty traits (within agreeableness), and lower order and orderliness (within conscientiousness) to this math and spatial cluster. CTCT offers an explanation. It identifies knowledge acquisition as one enabler of self-evolution, and visuospatial and quantitative abilities are both likely to be involved in self-evolution and thus related to ideas facet of openness. In contrast, traits involving politeness and modesty (within agreeableness) or order and orderliness (within conscientiousness) support maintenance of self in social environments and are therefore likely to drain energy from other intensive pursuits (e.g., negative relations with quantitative and visuospatial abilities). The focus of these traits on stability is counter to strategies focused on change, such as those involving visuospatial abilities and/or fluid abilities like induction.

Another example is Ackerman and Heggestad’s (1997) social trait complex, which included extraversion and emotional stability (i.e., low neuroticism) but no major cognitive abilities. Processing abilities (e.g., perceptual speed) could be motivationally relevant for detecting social cues and therefore useful as part of a social trait complex. The current meta-analyses investigated these “missing” personality traits and cognitive abilities, filling gaps in the nomological networks of hundreds of personality and cognitive ability constructs. The results highlight areas of overlap, suggest common causes/effects, and deepen understanding of prior research that omitted important personality or cognitive ability constructs when examining phenomena related to both domains. Even in personality-ability relations that have been examined previously, nearly half of the conclusions warranted revision, which has implications for practice as well as research.

Potential Limitations and Future Research

Despite significant strengths of current research, some potential limitations need to be addressed. We view several of these as providing fruitful avenues for future research.

Publication Bias

An often-cited threat to the validity of meta-analytic conclusions is publication bias. Several factors immunized the current meta-analyses against publication bias. First, and most importantly, 63% of the data analyzed here were derived from unpublished sources (e.g., data archives, unpublished conference manuscripts, dissertations). By including these materials, not only did we safeguard against publication bias, but we also brought to light and made accessible findings from otherwise ignored studies. Thus, as another benefit of this research, we are making coded data from these unpublished sources and newly computed relations from data archives available to the scientific community as part of the publicly available dataset associated with this research (see Appendix F).

Second, in addition to more than half of the effect sizes being derived from unpublished sources, few of the studies included in this research were conducted for the express purpose of examining personality and cognitive ability relations. Rather, these two sets of variables were often included in broader examinations of other phenomena such as physical health, educational attainment, job success, development and aging, interpersonal relations, and so forth. Relations between the variables we analyzed here were often incidentally reported in the correlation matrices of other works. That is, many of the identified and included sources were not intended to be direct examinations of personality-cognitive ability relations, reducing the threat of underreporting of null relations (see supplement of Wilmot & Ones [2019]).

Third, the threat of publication bias was also mitigated by the depth of the search strategies: multiple databases and archives were searched, relevant studies from reference lists of contributing research were culled, authors were contacted directly for missing metrics, and sources were not excluded because they were in different languages. In this same vein, the contribution of non-United States-based materials and data from nationally representative, large-scale studies1 were a further means of inoculation against publication bias.

Data Availability

Despite being based on the largest meta-analytic database ever amassed on the relations in question, for some cognitive ability-personality pairings, data were scant (e.g., fluid ability with persistence, verbal ability with cooperation, quantitative ability with intellect, visual processing with introspection). Although these gaps limit our ability to draw inferences about those relations, they also point to areas worthy of future research. We encourage individual differences researchers to proactively bring fresh data to bear on the associations for which there are few or no existing data.

Psychological research including personality and cognitive ability variables continues to be produced at an increasing rate. The series of meta-analyses presented in this research represent which relations have been observed between the two domains during a century of research. It is a comparative baseline for the next hundred years of research. Another potential area of expansion is sample age and type. We did not include children younger than 12 or clinical populations in our database, as we expected that personality-cognitive ability relations could be distorted in such samples (e.g., less differentiated, some relations more magnified or muted). Future research should systematically examine relations in these populations to shed light onto developmental questions and gain insights into psychopathologies.

Linear Associations

We acknowledge that the effect sizes examined in the present meta-analyses assume linear associations. This is because most primary studies assumed and reported zero-order, bivariate correlations. The possibility of non-linear relations has been raised (Ackerman, 2018). All personality traits are characterized by bipolarity (Krueger & Markon, 2006). Whether or not a given personality measure will assess extrema of a personality construct depends on the extremity of the items (Dilchert et al., 2014, 2019). Measures with extreme items (e.g., clinical measures) result in non-normal distributions and can thereby produce non-linear relations with normally distributed scores, such as those from standard cognitive ability tests. Future research should systematically examine whether measures assessing extrema in both cognitive ability and personality domains produce evidence of non-linear effects. While we took care not to include such effect sizes in our meta-analytic database, even if non-linear relations exist between personality and cognitive ability constructs, the linear associations detected and presented in this research would be underestimates of such non-linear relations.

Impact of Extremely Large Studies

Outliers are an important consideration in meta-analytic research since contributing studies with large sample sizes can influence results. Such weight is warranted because larger studies have smaller sampling error and are less likely to produce spurious associations. However, a quandary emerges if large samples yield findings differing from the remaining studies. For some of our meta-analyses, results varied when an extremely large study (Project Talent) was excluded. In these cases, Project Talent contributed data for approximately twice as many participants as any other primary study. As a result, the Project Talent data imparted considerable influence on some of the reported relations when N or K were comparatively small (i.e., there were few studies or smaller other samples and Project Talent was the largest sample contributor). Project Talent was designed as a nationally representative study of 1960s American high school students. If its data are not representative of the rest of the population of studies in terms of factors that impact the relations between personality and cognitive ability, then the results of otherwise small-N meta-analyses including them may be biased.

Although we had no a priori reason to think that the Project Talent data were deviant, all meta-analyses were run a second time excluding Project Talent. While some results did not change substantially (e.g., relations with cautiousness), others did. In most instances, across personality constructs and cognitive ability constructs, these exclusions led to relations shrinking in magnitude or shifting toward smaller or negative values. In general, the pattern of conclusions remained unchanged. There were two conscientiousness constructs for which inclusion of Project Talent data substantially altered conclusions: order2 and industriousness. When Project Talent data were included, these traits were more highly correlated with cognitive abilities, especially with acquired verbal abilities. The exclusion of Project Talent data also muted some of the effects reported for some extraversion facets, including those for activity. Relations with openness deflated the least (see Appendix E’s “Impact of Outlier Samples” section for further discussion). Tables presenting all the results without Project Talent data can be found in supplementary materials (Supplementary Tables 100–196 and 276–354 in Appendices H and J).

Causality and Longitudinal Effects

One major limitation of the present research is that we are not able to draw direct causal inferences due to the reliance on correlational data from cross-sectional investigations in nonexperimental settings. Longitudinal meta-analyses would supplement the current results and provide valuable insights for causal inferences (e.g., Luchetti et al., 2016). For example, if youth rank on industriousness is more predictive of adult rank on verbal ability than adult rank on orderliness, it might indicate that industriousness helps individuals accumulate knowledge or at least maintain verbal ability across the lifespan. Although such examinations would be an ideal extension of the current results, the number of effect sizes required to adequately fill a four-dimensional matrix of personality construct by cognitive ability construct by initial age by follow-up age is immense, and many cells would be sparsely populated (e.g., those with long age lags). Still, as data continue to accumulate from the multitude of longitudinal investigations, meta-analyses of longitudinal relations are recommended since the perspective afforded by such investigations would offer insights about the longer-term development of personality-cognitive ability relations, as well as personality traits and cognitive abilities separately.

Other Potential Moderators

For most of the 3,543 meta-analytic relations, the associated true standard deviations were nil, negligible, or very small. Therefore, for many personality-cognitive ability relations, there is little room for moderators to influence the strengths of relations (Schmidt & Hunter, 2014). In cases where there was room for moderators, age, gender, and cultural context may be fertile areas for future inquiry. One other potential moderator worth mentioning is the impact of stimulus type (e.g., “types of content” in Structure of Intellect model, Guilford, 1956). For example, when measuring quantitative ability, the stimulus could be a word problem or a geometric drawing.

Measurement Methods

In studying the past century of relations between personality traits and cognitive abilities, we were struck by how similar modern measures are to those used a hundred years ago. This is especially true for personality measures, which still consist mostly of people indicating how much certain words or phrases describe them. Such approaches require dedicated time to complete, are limited by a single perspective, are susceptible to respondents interpreting words differently, and are burdensome to collect. Digital technology, from wearable sensors (Wiernik et al., 2020) to passive digital exhaust (Hall & Matz, 2020), offers massive potential for overcoming the limitations of traditional measures and catapulting our understanding.

Distillation of Boundaries to Our Understanding

We compiled a multitude of diverse sources to examine the relations between personality traits and cognitive abilities. By including samples from more than fifty countries, research from the past century, measures from across disciplines, and samples from myriad demographics we were able to reveal patterns of personality-ability relations that were previously obscured. As a result, we were able to synthesize a theory that accounts for how these domains of individuality are related and why.

Despite investing over a dozen years into this project and striving to be thorough in our efforts, numerous avenues beckon for further research. Non-linear relations, moderators, and sub-group analyses are perennial opportunities, but the investigation of longitudinal trends is particularly tantalizing. Exploring the accumulated data across existing longitudinal studies would further elucidate how personality traits and cognitive abilities develop over time. Similarly, there is an opportunity to initiate and extend large-scale longitudinal studies with better measures of personality and ability than ever before. Finally, there is a significant opportunity to better measure the environments of individuals alongside their traits, abilities, and goals.

References

Ackerman, P. L. (2018). The search for personality–intelligence relations: Methodological and conceptual issues. Journal of Intelligence, 6(1), 2.

Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality, and interests: Evidence for overlapping traits. Psychological Bulletin, 121, 219–245.

Dilchert, S., Ones, D. S., & Krueger, R. F. (2014). Maladaptive personality constructs, measures, and work behaviors. Industrial and Organizational Psychology, 7(1), 98–110.

Dilchert, S., Ones, D. S., & Krueger, R. F. (2019). Personality assessment for work: Legal, IO, and clinical perspective. Industrial and Organizational Psychology, 12(2), 143–150.

Guilford, J. P. (1956). The structure of intellect. Psychological Bulletin, 53(4), 267.

Hall, A. N., & Matz, S. C. (2020). Targeting item–level nuances leads to small but robust improvements in personality prediction from digital footprints. European Journal of Personality, 34(5), 873–884.

Hedges, L. V., & Nowell, A. (1995). Sex differences in mental test scores, variability, and numbers of high-scoring individuals. Science, 269(5220), 41–45.

Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comorbidity: A model-based approach to understanding and classifying psychopathology. Annual Review of Clinical Psychology, 2, 111–133.

Luchetti, M., Terracciano, A., Stephan, Y., & Sutin, A. R. (2016). Personality and cognitive decline in older adults: Data from a longitudinal sample and meta-analysis. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 71(4), 591–601.

Schmidt, F. L., & Hunter, J. E. (2014). Methods of Meta-Analysis: Correcting Error and Bias in Research Findings (3rd ed.). Sage Publications.

Wiernik, B. M., Ones, D. S., Marlin, B. M., Giordano, C., Dilchert, S., Mercado, B. K., Stanek, K. C., Birkland, A., Wang, Y., Ellis, B., Yazar, Y., Kostal, J. W., Kumar, S., Hnat, T., Ertin, E., Sano, A., Ganesan, D. K., Choudhoury, T., & al’Absi, M. (2020). Using Mobile Sensors to Study Personality Dynamics. European Journal of Psychological Assessment.

Wilmot, M. P., & Ones, D. S. (2019). A century of research on conscientiousness at work. Proceedings of the National Academy of Sciences, 116(46), 23004–23010.